Responsible AI Policy

1. Purpose

This policy outlines our commitment to responsible AI use in healthcare, ensuring compliance with AHPRA regulations and ethical guidelines. Our AI systems support healthcare professionals and business operations but do not replace clinical judgment, medical decision-making, or regulatory responsibilities.

‍

2. Compliance with AHPRA and Regulatory Standards

• Our AI solutions align with Australian Health Practitioner Regulation Agency (AHPRA) requirements and other relevant Australian healthcare regulations.

• AI-assisted features in DappleOS™ are designed solely for efficiency and time-saving, not for clinical decision-making or patient care recommendations.

• We actively review AI implementations to ensure they remain within compliance boundaries and do not constitute a regulated medical device under the Therapeutic Goods Administration (TGA).

‍

3. AI is Not a Replacement for Healthcare Professionals

• AI in DappleOS™ is an assistive tool designed to streamline workflows and improve efficiency, but it does not provide clinical recommendations.

• AI is used to support clinicians with:

• Treatment planning (e.g., structuring appointments and follow-ups)

• Patient summaries (e.g., condensing key information from records)

• Response suggestions (e.g., assisting with communication templates)

• Transcription (e.g., real-time note-taking with patient consent)

• Drafting clinical notes (e.g., summarising consultations for review)

• AI-generated content must be reviewed and validated by a healthcare professional before any action is taken.

• AI cannot provide medical diagnoses, prescribe medication, or make autonomous clinical decisions.

‍

4. AI is Not a Medical Device

• AI tools in DappleOS™ are not classified as a medical device under the TGA’s Medical Device Framework.

• No AI-powered clinical recommendations are generated, and AI does not function as a diagnostic or decision-making system.

• AI is strictly used for administrative efficiency, including documentation support, transcription, and workflow automation.

‍

5. AI in Non-Clinical Applications

In addition to assisting clinicians, AI is also used for non-clinical activities within DappleOS™, including:

• Marketing content review and optimisation (e.g., refining messaging and engagement strategies)

• Operational efficiency improvements (e.g., automation of workflows)

• Optimisation recommendations (e.g., resource allocation and performance insights)

• Reporting insights (e.g., analysing data for trends and decision-making support)

• Customer support assistance (e.g., suggested responses for service teams)

AI in these areas is used to enhance business operations while ensuring full compliance with ethical and legal standards.

‍

6. AI Ambient Transcription & Patient Consent

• AI-powered ambient transcription requires explicit patient consent per AHPRA guidelines before activation.

• No audio recordings are stored—transcription is performed in memory only, meaning no audio data is retained after processing.

• Chunked transcription is used to mitigate risks, ensuring real-time, ephemeral processing rather than continuous recording or storage.

• Patients and healthcare providers are clearly informed when AI transcription is in use.

‍

7. Data Privacy & Model Security

• DappleOS™ does not use input or output data to train models.

• Model invocation communications stay within the DappleOS™ network to ensure data privacy and security.

• No patient or user data is used to improve AI responses or refine system models.

• AI-generated outputs are based on predefined algorithms and structured rules, ensuring no continuous learning or adaptive behavior that could introduce compliance risks.

• Data security is a priority, and we do not retain or process user interactions beyond their immediate session.

‍

8. Changes to AI Models and Outputs

• If DappleOS™ modifies the prompts, methodologies, or approaches used to generate AI outputs, we will inform users in advance.

• Any changes that affect AI-generated outputs will be clearly communicated to ensure users are aware of potential differences in functionality or behavior.

• Users will be given guidance on how updates may impact workflows and any necessary adjustments to their usage of AI-assisted features.

‍

9. Transparency and Informed Use

• Any AI-generated content or assistance is clearly disclosed to users (both clinicians and patients) at the point of interaction.

• AI outputs are positioned as informational and administrative, with no clinical recommendations provided.

• Patients, healthcare providers, and staff are always made aware of when and how AI is being used within DappleOS™.

‍

10. Guardrails, Abuse Prevention & AI Response Tempering

• Guardrails are implemented to prevent misuse, ensure ethical interactions, and restrict inappropriate AI outputs.

• AI response temperature is set to low, prioritising factual accuracy and controlled responses over creative or speculative outputs.

• AI interactions are monitored for abuse, and safeguards are in place to mitigate potential harm, bias, or misinformation.

‍

11. Continuous Monitoring and Ethical Oversight

• We regularly audit our AI models to assess fairness, accuracy, and alignment with ethical healthcare and business practices.

• Bias mitigation strategies are applied to prevent discriminatory or misleading AI-generated outputs.

• Our Clinical Advisory Committee oversees responsible AI implementation and ensures adherence to healthcare safety, regulatory, and privacy standards.

‍

12. Accountability & Feedback

• AI use within DappleOS™ is subject to human oversight at all times.

• Healthcare professionals remain fully responsible for their clinical decisions, even when using AI-assisted features.

• We encourage user feedback to continually refine AI capabilities while ensuring compliance with evolving AHPRA guidelines and ethical AI practices.

‍

13. Review & Updates

This policy will be regularly reviewed and updated in response to regulatory changes, technological advancements, and ethical considerations to ensure responsible AI use in healthcare and business operations.